December 5, 2013
I recently had a meeting with executives from Tech Mahindra, an Indian-based IT services company, which was refreshing for the both the candor with which they discussed the overall mechanics of a support and integration model with significant components located half a world away, as well as their insights on the realities and limitations of automation, one of the hottest topics in IT operations today.
On the subject of the mechanics and process behind their global integration process, the eye opener for me was the depth of internal process behind the engagements. The common (possibly only common in my mind since I have had less exposure to these companies than some of my peers) mindset of “develop the specs, send them off and receive code back” is no longer even remotely possible. To perform a successful complex integration project takes a reliable set of processes that can link the efforts of the approximately 20 – 40% of the staff on-site with the client with the supporting teams back in India. Plus a massive investment in project management, development frameworks, and collaboration tools, a hallmark of all of the successful Indian service providers.
From a the client I&O group perspective, the relationship between the outsourcer and internal groups becomes much more than an arms-length process, but rather a tightly integrated team in which the main visible differentiator is who pays their salary rather than any strict team, task or function boundary. For the integrator, this is a strong positive, since it makes it difficult for the client to disengage, and gives the teams early knowledge of changes and new project opportunities. From the client side there are drawbacks and benefits – disengagement is difficult, but knowledge transfer is tightly integrated and efficient.
On the topic of automation, the Tech Mahindra group was very forthcoming with some anecdotal data that should be required reading for any IT executive planning to engage in major automation efforts as part of their infrastructure maturity process. Tech Mahindra has made major investments in their internal tools and frameworks for automating customer IT and BT processes. These tools were developed to help them operate more efficiently, i.e. to make more money, and are constantly being revaluated against the inherent Darwinism of private enterprise – do they add to the bottom line or not. Tech Mahindra has been working on these tools for the last five years, and has developed automation templates for approximately 1400 business and IT processes to date. All well and good, but how effective are these automation capabilities in the real world. The answer was in one sense disappointing, in another very consistent with industrial reality.
In a word, automation is much less automated than its conceptual proponents would have you believe. The Tech Mahindra team said that somewhere between 60 and 80 percent of their automated templates needed customization to account for variations in the customer’s business process and infrastructure. Additionally, the discovery of and documentation of the business processes cannot be automated, but requires dedicated effort on both the part of Tech Mahindra and the clients. At Forrester we have long preached that the biggest limitations in the process of automating any substantial subset of IT operations is usually the client process itself, but it is always nice to have to confirmed by professionalsJ Another interesting bump in the road to successful implementation is that fact that this process diligence often discovers that for many operations their often are no formal processes, which means that they have to be developed in-flight.
So, is Tech Mahindra shining a light on a previously unseen, or at least ignored by common consent aspect of large integration projects, or are they simply not very smart and way behind the industry technology curve? All of Forrester’s evidence points to the former, and that they are to be applauded for their candor and willingness to educate. The simple fact is that the majority of companies which have entered the market with exotic automation (or orchestration as it has been called) products have failed. Many of them succeeded in getting a large marquee account and then stalled, probably because they died under the weight of actually implementing the product after an impressive POC of limited scope and carefully controlled complexity. Others have been acquired, but for all the “sturm und drung” of automation, orchestration, adaptive systems, self-organizing software, etc., there is little sign of their passing. Except for the very largest web companies, whose existence depends on successful automation, there is little evidence of wide-spread process automation in enterprise IT beyond mass rollout of patches and updates, and network-wide backups (and these often have regular failures requiring manual intervention). The truth appears to be that it is a grinding and difficult process, and that it can be made to work only with significant effort and a lot of up front work.
In effect, automation is not in any sense itself an automated process. But is it worth it? It depends on your scale and your expectations. Tech Mahindra shared details of one of their largest and oldest contracts, currently in excess of 8 years and still going with a major North American telecom provider with global operations, for whom Satyim performs all of their infrastructure management. The company requires a support team of approximately 1500 Tech Mahindra employees who manage approximately 75,000 servers, 45,000 network devices and 25 PB of online data. The staff is about 40% resident in company facilities and the remainder in remote locations. Since this was one of Tech Mahindra’s earliest clients, it has not been implemented using the automation framework tools, and Tech Mahindra is in the early stages of introducing automation to many of the internal customer processes. In all, Tech Mahindra stated that they expect to gain maybe an additional 20% efficiency improvement in infrastructure operations – not a huge number against early claims made by an enthusiastic industry, but actually very impressive in terms of the actual dollar cost savings in an operation this size which probably has an amazingly complex set of processes.
All in all, an eye-opening meeting with a company that clearly understands its market and clients and is not afraid to communicate the good and the not so good about their industry.